Fatigue Assessment
Fatigue Assessment
Fatigue Assessment
Summary
Methodology for fatigue life assessment is discussed. Fatigue life dependants for the gas lift lines
are described. Actual parameters for the calculations are identified. Currently missing input data
and its replacement data are highlighted.
References:
/1/ PD 5500 !00" #pecification for $nfired fusion %elded pressure &essels'
Anne( C Assessment of &essels sub)ect to fatigue
/!/ *+ 1",,5-" !00! $nfired pressure &essels . Part " Design
Chapter 1/ Detailed assessment of fatigue life
/"/ D+0-1P-D101 !00/ 1ecommended Practice #tructural Analysis of Piping #ystems
Chapter ".1! Fatigue calculations2 App. 3 Fatigue Calculation *(.
/,/ P14M*C5 *(tended fatigue assessment as done on 6reater *7ofis7 assets
/5/ A.A8MA1-+9## Fatigue 5andboo7 4ffshore #teel #tructures'
:apir' :rondheim' 1;/5
Page 1 of ;
Introduction
Fatigue is a failure mode that %ill occur under se&ere cyclic conditions' e.g. a pipeline sub)ected to
%a&es. <f a large enough number of load fluctuations are allo%ed then fracture %ill e&entually
occur. :he load itself may not be large enough to cause immediate failure' but it is the number of
load fluctuations that causes failure.
Fatigue failure %ill occur at the %ea7est part of a piping system %ere the stress le&el is highest.
=elds are normally %ea7er than smooth pipe since there %ill al%ays be small imperfections or
fractures in any %eld. :he si>e of these imperfections relates to the %eld ?uality. <f the cyclic stress
range goes abo&e the lo%er threshold for fatigue these small imperfections or crac7s %ill start to
propagate/gro% and may reach a critical si>e ending in fatigue failure.
Damage may appear for e&ery load cycle' and the damage accumulates. =hen the accumulated
damage has reached a critical le&el @critical crac7 si>eA failure %ill occur.
:he most important component under fluctuating stress and strain is called the stress range @or
amplitudeA. A stress range is the algebraic sum bet%een a pea7 stress and its subse?uent &alley.
4ne method to calculate fatigue life is based on the #+ Miner-Palmgren approach. 5ere the cyclic
stress range' #' is related to the number of cycles to failure' +. #+-cur&es ha&e been made on the
basis of laboratory tests. <n %elded piping systems' the %eld constitutes the %ea7 lin7 considering
fatigue strength. Fatigue design of piping is therefore based on #+ data obtained from realistic
%elded specimens. :he #+-cur&e for metals has the form
a N S
m
r
=
%here S
r
is the cyclic stress range' N is the number of cycles to failure' %hereas m and a are
constants.
:he m and a constants represent the different fatigue cur&es and the &alues depend on %hich %eld
detail is under consideration.
For further bac7ground information please see belo% sections Extended fatigue assessment
/4/
and
Fatigue analysis of wellhead flowlines
/3/
.
:here are "0 gas lift lines to be installed on 0alhall Flan7 +orth and #outh' 15 on each platform.
All gas lift lines are made of !B seamless CMo pipe.
<n order to estimate the fatigue life of these lines one has to address the fatigue life dependants.
:he follo%ing dependents are identified based on the reference documentation listed belo%.
A. Fatigue Cur&e
D. <mposed Displacements
C. +umber of Cycles
D. Piping 1estraints
*. Pipe Material
Page ! of ;
A. Fatigue Curve
:he m and a constants depict %hich fatigue cur&e to be used. :hey are dependant on %hich %eld
detail is under consideration as %ell as ho% the %eld is tested. :here are t%o %elds details that are
applicable for the 0alhall Flan7 gas lift lines.
- According to *+ 1",,5-" table 1/-, they are detail no. E.1 Full penetration butt %elded
nec7 flange and detail no. 1.5 Full penetration butt %elds. :he fatigue class is decided
based on %hat 7ind of testing is performed. <f testing group 1 or ! is applicable then fatigue
class C" may be used as long as full penetration is assured' if testing group " is applicable
then fatigue class ,0 is to be used. :he lo%er fatigue class number used results in the
shorter calculated fatigue life.
- According to PD 5500 fatigue class * is applicable for the t%o %eld details as long as full
penetration can be assured as %ell as the %eld is pro&ed free from significant defects by
+D: testing.
<n order to use the correct fatigue cur&e one needs to 7no% the %eld detail' %elding method and
ho% it is tested.
According to EN 13445-3
For the Valhall Flank gas lift lines fatigue cure 4! is conseratiely chosen and the
sti"ulated m and a constants are#
m $ 3 a $ 1%&' x 1!
11
for stress range cycles 1!
&
(N(5x1!
)
m $ 5 a $ 1%11 x 1!
14
for stress range cycles 5x1!
)
(N(1!
'
According to *+ 55!!
For the Valhall Flank gas lift lines fatigue cure E is chosen and the sti"ulated m and a
constants are#
m $ 3 a $ 1%!4 x 1!
1&
for stress range cycles N(1!
,
m $ 5 a $ &%&- x 1!
15
for stress range cycles N.1!
,
B. Imposed Displacements
All loads %ith a cyclic nature are to be considered for the gas lift lines. :he piping %ill be sub)ect
to pressure cycles' thermal cycles and forced displacement cycles %hich stem from ocean %a&es.
:he %a&es deflect the riser conductor gi&ing hori>ontal mo&ements of the %ellhead @(-mas treeA.
:he ma)or contributor for the fatigue life for gas lift lines are the %ellhead hori>ontal
displacements. :he effect of the %a&es on the riser pipe in relation to hori>ontal %ellhead
mo&ements depend on the supporting of the conductor. Conductor centrali>ers are normally
installed in order to limit the shac7ing of the %ellheads. :hese should be tight in order to limit the
hori>ontal mo&ements as much as possible. Any reduction of %ellhead hori>ontal displacement
%ill reduce the cyclic stress range and subse?uently affect the fatigue life significantly.
Conductor analysis is normally performed %here the hori>ontal displacements of the %ellhead may
be found.
For the 0alhall Flan7s the only source for the hori>ontal deflections are found on the e(isting stress
isometrics done by 5eerema :Fnsberg. 5ere 10mm hori>ontal deflections are stipulated.
/ence% 1! mm is used for ' a""roach directions of the waes N% 0% 1% E% 01% NE% 0E% N12
Page " of ;
C. Number of Cycles
Another ma)or contributor to the fatigue life is the number of cycles for each stress range. :he
number of hori>ontal displacements for the %ellhead and subse?uently the gas lift lines is related
to the number of %a&es acting on the riser conductor. =a&e occurrence diagrams are used for
finding the number of cycles anticipated to act on the riser conductors. :hese are usually found in
en&ironmental or oceanographic design data reports for the oil field in ?uestion. :he 0alhall field
is in the &icinity of the 6reater *7ofis7 area as e(plained in report no. #/*P:/0E"/0, 3uly !00E
0alhall Field 4ceanographic and Meteorological Design Data #ummary. :he *7ofis7
en&ironmental data chapter "." Fatigue =a&es sho%s all %a&e occurrences o&er a "0 year period
from / approach directions in table "."-1. #ee section Fatigue 1aes belo%.
An e(treme conser&ati&e approach to fatigue life calculations %ould be to consider all %a&es'
regardless of %a&e height' to deflect the %ellhead the full 10mm hori>ontal mo&ement. :his is not
implemented in the 0alhall Flan7 gas lift fatigue calculations.
A conductor analysis %ould sho% %hich %a&e heights are to be ta7en into consideration. A
conductor analysis for the gas lift %ellheads are not currently found so an analysis for an *7ofis7
!/,M riser is used' ref. page 1" and "; of doc. 1iser Analysis' ConocoPhillips *7ofis7 <<
De&elopment Pro)ect' !/,M. 50,!E!;' 1e& +C.
For this riser it is considered that only %a&e heights ,m and abo&e has impact on the fatigue life.
For this reason only %a&e occurrences for %a&es 5G, and abo&e are considered in the fatigue life
calculations for the 0alhall gas lift lines. :he follo%ing table is used
Table 3.3-1 adjusted to include only waves 4m and above
H T
OMNI N N S S S! ! N!
(m) (s)
3 14!3 1 " " " " " " " 1
14! 1 " " " " " " " 1
1 14!1 4 " " " " " " 1 3
" 13!# 1$ 3 " " " " " 4 #
1# 13!% $# & & 4 " " " 1% 3'
1' 13!& 14" 11 % 4 4 4 4 33 %3
1% 13!3 "% 1% 1 % & & &" 1"#
1$ 13!1 3$# 31 1 1 $ 1 ' '# 1#"
1& 1!# &&& 44 3" 1 1" 1 14 13' %%
14 1!% 1 "4 #1 $1 &1 % $" 34 311 &$#
13 1!& 1 '"$ 1% '$ '' &" 11& $& 4'3 %#
1 1!3 3 $14 3& 1&' "$ 1& #& 1%% ##& 1 43
11 1!" 1" '"1 $3# 4% %1# 4' 1 "## %4% 3 "14 3 $%4
1" 11!% #31 1 # '11 1 %'# 1 '& %%4 1% $ 3%% $ 4#4
# 11!3 4# 1$4 4"$ 1 &&3 4 4# 3 333 $ $$1 & ### 13 443 11 %%
' 1"!' 1" $3" & &11 3 43' 1 '' # 4'& 1% 1'% 1% '3# 3 1'4 1&'
% 1"! &3 %$' 11 1#& $ 431 3 11% 3 34' 3& 344 4$ 434 $4 '"$ 34 "#3
$ #!$ $11 1" % #'$ 1& "1" '" #4 $" 1$& '1 31 13 1"1 1&1 33& %1 3$'
& #!" 1 &% ''$ %& #& 3% $' " "3 1&# '#3 1#4 "'" 33 '11 3$4 4#% 1$" #%&
4 '!& 3 #$" &"" 4 '#3 1" 41& 4#% &%# 43% $&# 4'& $3" #1 $1 #"3 "3 3#$ $'#
TOT() $ &$4 '%$ 3&" 34' 1$' 14% '3 '$4 $#& '%4 '4 $"' 1 44 " 1 &4" '"" %1" 13
3he actual fatigue wae data for the Valhall Flank "latforms should 4e "roided2
Also "roided should 4e riser analyses showing the magnitude of hori5ontal wellhead
deflection as well as which wae height occurrences to consider for calculating the gas lift line
fatigue life2
Page , of ;
D. "iping Restraints
:he supporting of the gas lift lines closest to the %ellhead depicts the stress range. :he more
restraints there are the higher the stress range %ill be as the fle(ibility of the piping is reduced. :he
restraints also affect the natural fre?uency of piping systems. A natural fre?uency abo&e , 5> is an
indication that the piping is ade?uately supported. For piping connected to %ellheads this goal is
often not obtained since it directly affects the fatigue lifetime.
For the gas lift lines the su""orting is designed to kee" an acce"ta4le fatigue life oer 3!
years and the natural fre6uency as far as "ossi4le oer 4/52 For some gas lift lines the
natural fre6uency o47ectie is not reached2 8t is considered more im"ortant to kee" the
fatigue life acce"ta4le2 3he fatigue life and the lowest natural fre6uency results are to 4e
sti"ulated on the stress sketch for each gas lift line2 Friction effects are included as a friction
factor of !23 is included for the guide su""orts closest to the wells2
. "ipe #aterial
:he #+-cur&es are related to !!mm thic7 pipe and an *-modulus e?ual to !0;000 MPa. :he stress
range must be corrected for other thic7nesses and other materials. 0alues belo% !!mm and &alues
abo&e !0;00 MPa are not permitted in the calculations. For further bac7ground see belo% section
Extended fatigue assessment
/4/
.
3he a""lica4le "i"e wall thickness is '2,4mm for the gas lift lines2
3he E-modulus used is for am4ient condition and is &!!!!! 9*a2 3his is conseratie since
the o"erating tem"erature for the gas lift lines is -&5 :;2
Page 5 of ;
$tended fatigue assessment
/4/
Theoretical background
=a&e loadings at the conductors initiate displacements and rotations at the H-mas trees. #ince the number
of complete load cycles in the H-mas tree mo&ement case is %ell abo&e E000' an additional fatigue chec7
load case must be in&estigated. :his may be done according to pro&isions in PD5500' I#pecification for
unfired fusion %elded pressure &esselsB' anne( C.
:he fatigue strength of a unit is usually go&erned by the fatigue strength of details as e.g. %elds and its
fatigue strength is assessed on the basis of #-+ cur&es in %hich the repeated stress range #r is plotted
against the number of cycles to failure' +.
:he #-+ cur&es for assessment of %eld details ha&e the follo%ing form
A N S
m
r
=
%ith
+ number of cycles to failure
#r repeats stress range
m constant depending on fatigue class and number of cycles
A constant depending on fatigue class and number of cycles
:he e?uations are deri&ed from fatigue tests based on specimen of ma(imum !! mm thic7ness and an *-
module of !.0;J10
5
MPa. :a7ing into account different material and plate thic7ness' the follo%ing e?uation
is deri&ed
m
r
m
E
S
e
A N
=
5
, /
10 0; . !
1 !!
%ith
e ma(imum of greatest thic7ness including corrosion allo%ance or !! mm
* Modulus of elasticity
Fatigue damage can according to the Miner Palmgren e?uation' be e&aluated by the cumulati&e effect of
different types of stress cycles' %here the follo%ing condition shall be met
1
i
i
N
n
%ith
n
i
number of times each stress cycle' #
ri
' %ill occur during the life of the structure
+
i
fatigue life @number of cycles to failureA corresponding to #
ri
' obtained from
appropriate #-+ cur&e
According to the design code' the fatigue assessment shall be based on the primary plus secondary stress
category and the full stress range is to be used.
Analysis approach
Page C of ;
4n most installations' conductor analysis has been underta7en. Displacements and rotations at the location
of the (-mas tree' %here flo% lines and gas lift lines are connected' are determined based on a range of
significant %a&e heights and associated number of occurrences for / approach directions of the %a&es.
#ince the &ertical displacements at the (-mas tree due to %a&e motion are negligible' the hori>ontal
mo&ements are basis for the fatigue analysis.
Depending on the si>e of the calculated displacements' the %a&e heights are classified in bloc7s and the
total number of cycles' n' is determined for the four main directions2 +orth-#outh' *ast-=est' +ortheast-
#outh%est and +orth%est-#outheast.
#tress analyses are performed based on the ma(imum deflections found in each bloc7 applied in the
considered four main directions. :he corresponding stress range is found based on the calculated Caesar <<
code stress @s
code
G s?rt@#
b
!
K ,#
t
!
A. :he total number of cycles to failure + and the fatigue damage are for
each bloc7 and main direction determined according to PD 5500 using fatigue class * @all %elds sub)ected
to +D: and fault freeA.
Frictional effects in supports may be of significance in the fatigue analysis cases and must be e&aluated as
they tend to increase the stiffness of the piping system related to the applied (-mas tree deformations.
Production flo% lines must be chec7ed for flo% induced fatigue loads' but this does not apply to gas lift
lines as they are e(posed to one-phase flo% only.
Page E of ;
Fatigue analysis of %ell&ead flo%lines
/3/
=ellhead flo%lines must be sub)ected to a comprehensi&e fatigue analysis gi&en the fact that they are
e(posed to high cycle loadings from Hmas tree mo&ements and flo% induced &ibrations.
As A#M* D "1." only to a limited degree ta7es into account fatigue damage' a more detailed fatigue
calculation according to PD5500 should be performed for all flo%lines and gas lift lines.
=a&e loadings at the conductors initiate cyclic loads at the Hmas trees and an additional fatigue chec7
should be performed.
:he flo% induced loads %ith the largest contribution to the fatigue life are not the design slug loads' but
rather loads generated by minor density fluctuations in the %ell stream. :hese flo% induced loads are not
applicable for gas lift or %ater in)ection lines.
According to PD5500' the fatigue assessment shall be based on the primary plus secondary stress category
and the full stress range is to be used.
:he design lifetime of flo%lines and gas lift lines should be "0 years unless other%ise specified by
Company. Fre?uent inspection should be initiated after 1/" of the estimated lifetime.
Frictional effects in pipe supports may be significant in fatigue analysis since they tend to increase the
system resistance to Hmas tree mo&ements' ref *+1",/0 1!.!.10.".1. :hese effects tend to be of
importance in systems %here there are one or more supports @not spring onlyA relati&ely close to the Hmas
tree and %here no line stops pre&ents the lines from mo&ing relati&ely to the other supports.
=hen preliminary e&aluations indicate that frictional effects may be of significance' these effects shall be
more thoroughly in&estigated through incorporation in the fatigue load cases.
<n order to do a representati&e analysis' the analyst should bear in mind that the fatigue %a&e loads @Hmas
tree mo&ementsA represents oscillations around a steady state deformation pattern created by %ell gro%th'
%eight' temperature and pressure. 5ence' the friction loads should be based upon the reaction loads from
these steady state solutions.
:he results from fatigue analysis should be reported in the pipe stress report. Any spool not meeting the "0
years fatigue life design re?uirement shall be identified. Design life time is to be stated on the #tress
<someric Dra%ing for the pipe.
Page / of ;
:he follo%ing is assumed to be ta7en from the 6reater *7ofis7 en&ironmental data
'.' Fatigue !aves
:he preferred fatigue analysis is the deterministic fatigue analysis described in 1ef. LCM. :he follo%ing %a&e
occurrence diagram o&er a "0 year period sho%n in :able "."-1 may be used. :hese &alues may also be used if the
stochastic fatigue analysis is applied.
3a4le
323-1
<ndi&idual %a&e height' 5@mA &ersus chart north %a&e direction @%a&es coming fromA for a period of "0 years.
5
@mA
:
@sA
4M+< + +* * #* # #= = +=
!" 1,." 1 0 0 0 0 0 0 0 1
!! 1,.! 1 0 0 0 0 0 0 0 1
!1 1,.1 , 0 0 0 0 0 0 1 "
!0 1".; 1C " 0 0 0 0 0 , ;
1; 1".E C; 5 5 , 0 0 0 1E "/
1/ 1".5 1"; 11 E , , , , "" E"
1E 1"." !0/ 1E 1! E ! 5 5 50 10;
1C 1".1 "E1 "1 !1 1! C 1! / /; 1;0
15 1!.; 55C ,, "0 !1 10 !1 1, 1"/ !EE
1, 1!.E 1!0, ;1 C1 51 !E C0 ", "11 5C;
1" 1!.5 1/0E 1!E /C // 50 115 C5 ,/" E;!
1! 1!." "C1" !"5 15/ !0C 1!5 !;5 1EE ;;5 1,!"
11 1!.0 10/01 C"; ,!E E1; ,/! 10;; E,E "01, "CE,
10 11.E !!;"1 1!!; /11 1E/; 1!/5 !EE, !1E! C"EE C,;,
; 11." ,;1C, !,0C 155" ,,;! """" CCC1 5;;; 1",," 11!EE
/ 10./ 1!0C"0 5511 ","/ 1!/!/ ;,/5 1E1/E 1E/"; "!1/, !!15/
E 10.! !5"EC/ 111;5 C,"1 "!11E !"",/ "5",, ,C,", C,/0C ",0;"
C ;.C C11!10 !E;/C 15010 /0;!, C01C5 /1"!1 1!"101 151""5 E1"C/
5 ;.0 15!E//E E5;!5 "EC/! !0!0!" 15;/;" 1;,0/0 ""!/11 "C,,;E 1C0;E5
, /.5 ";C0,;; !!,/;" 10!,15 ,;E5E; ,"EC5; ,/5C"0 ;1!C1! ;0"0!" ";CC/;
" E./ 10C,!";/ E","E1 "0551, 1!0;/E/ 1!15CEE 1!;;5,/ !50C""! !!;5CE0 10E5,0;
! C.E "00";"C, !CE0C/" 1050"/E !;55151 ""/C0;! "E;,1,! C//"5!C C0!51CE "!E,!1E
1 5.0 //,CEC!; ;5C1!10 "EC/E/; EC,,!E" ;5!/,EC 115/"/10 1;1EEC0; 1CC!5/5" 105EEC0/
0.!5 !.; 55C,5;;" C01";/C !"E05C! ,/0/!"5 5;;"";C E!/C1;E 1!0C!C/5 10,5EC", CC5"!;;
:4:A
8
1;1"C0!C5 1;""05;/ ECC",0! 1E,50,0! !0/1;51, !,E//"05 ,!0E!1E5 "C;,51!, !!!;0E,5
*(ample 4&er a period of "0 years one %ill on the a&erage e(pect to encounter 5';;; indi&idual %a&es from #= %ith
/.5 m N 5 N ;.5 m @central &alue 5 G ; mA. :he a&erage period of these %a&es may be ta7en as : G 11." s.
Page ; of ;